Redundancy in ATM systems is currently provided using multiple Permanent Virtual Channels (PVCs), and uses a Network Management System (NMS) to detect failure on a primary path and to switch to the redundant path. The NMS detects failure both within the network at network-network interfaces (NNIs) and at customer attachment points at User-Network Interfaces (UNIs). Not only does this approach use the inefficiencies of multiple PVCs with their fixed routing, but a heavy load is also placed on the NMS.
Moving control of multiple endpoint (MEP) redundant connections from the NMS to within the network would allow drastically faster fault detection and recovery, as well as greater scalability.